Reflections on Lynx Tech Blog

Lynxgame: A shelved mini-game factory plan

Mon, 10 Aug 2026 12:00:00 +0800

Origin Qbitai ran an article saying that someone used Opus 5 with a single 2,000-word prompt to generate a playable speedboat racing game called INK TIDE—built with Vite + TypeScript + Three.js, with all assets and code generated—and that someone else recreated it with GPT-5.6 Sol for just $5. My first reaction after reading it wasn’t “this game is really polished,” but rather: “AI has already made it realistic to watch which game is trending and clone one in a day.” If the production barrier ha

274 Commits in a Month: I Shut Down More Projects Than I Started

Mon, 10 Aug 2026 10:00:00 +0800

First, Let’s Put the Numbers on the Table Over the past 31 days, I sampled a dozen or so repositories I have on hand: 274 commits. The most active one, lynxhot, had commits on 21 out of those 31 days. A blog that was only set up on August 3 racked up 100 commits in 8 days. On top of that, there are still more than a dozen research reports sitting on my hard drive—from Tesla’s open-source approach to car building to layered trash bags, from a content analysis of 371 Zhihu answers to the whereabou

Prompts, Context, Harness: How the Focus in AI Has Shifted Three Times Over the Past Four Years

Mon, 10 Aug 2026 04:30:00 +0800

First, an Experiment In early 2026, developer Can Bölük ran an experiment: same model, same set of tasks, nothing changed except the engineering layer around the model—specifically, the format used by the harness to handle code patches. The task success rate jumped from 6.7% to 68.3%. Tenfold. Not a single line of the model changed. That number has been making the rounds in AI circles lately because it turns something many people had vaguely sensed into a conclusion that is impossible to ignore:

It Won't Do It Unless I Ask: My First Few Days with deepseek-v4-flash

Mon, 10 Aug 2026 00:00:00 +0800

If I Don’t Ask, It Doesn’t Act I recently switched to a new model for work: deepseek-v4-flash-0731. After using it for a few days, I’ve figured out its temperament: if I don’t ask, it doesn’t act. Tell it to fix a bug, and it fixes only that bug. Three lines away there might be an obviously explosive bit of code, but if I don’t call it out, it won’t touch it. Once it’s done, it won’t ask a follow-up either—things like “Want me to add a test while I’m here?” are simply not in its vocabulary.

Handmade: a little private corner of this blog

Fri, 07 Aug 2026 23:52:00 +0800

Why I’m Setting Aside This Little Plot of Land The truth is, most of the posts on this blog were written with help from AI. I even admitted as much in the post “So Efficient It Doesn’t Feel Like Me”: gathering information, organizing ideas, even drafting the final piece—AI is absurdly fast at all of it, and I’ve long since become unable to do without it. But precisely because I can’t do without it, I’ve found myself wanting more and more to leave one place untouched by it.

To Embrace AI Is to Embrace It Fully

Fri, 07 Aug 2026 23:11:00 +0800

Some People Don’t Like AI-Generated Stuff Some people around me are pretty resistant to AI-generated content. Their reasons usually fall into a few buckets: it feels too machine-made, fake at a glance; it wasn’t written by the person themselves, so it has no soul; asking AI to write for you is basically admitting you can’t write. I get it, but I don’t agree. My Stance: If You Embrace It, Embrace It Fully My view is simple: if you’re going to embrace AI, then embrace AI all the way.

So Efficient It Doesn't Feel Like Me: A Moment of Self-Reflection on This Blog

Fri, 07 Aug 2026 23:09:00 +0800

Efficiency Is Sometimes the Problem Here’s something a bit awkward: when I write this blog, a lot of the drafts are actually organized for me by AI. It’s not laziness—it’s just that it’s genuinely fast. I throw in a rambling mess of thoughts, a half-formed idea, an article I found interesting, and it churns out a fully structured draft in seconds. It also auto-translates into English, auto-publishes, and auto-backs up. The efficiency is staggering.

An Investment Answer I Read Three Times: Right, Wrong, and Driving by the Rearview Mirror

Fri, 07 Aug 2026 22:51:00 +0800

Zhihu Taught Me a Lesson I recently came across an answer on Zhihu, and after reading it I felt a bit struck: the quality of discussion in that community seems a lot higher than I remembered—or maybe it was always high, and I just used to keep running into the low-quality part of it. The author of the answer is Simon, and the original is here. Using the question “Why can’t most people achieve a major breakthrough?” as a starting point, he explained how he discovered an investment blogger worth f

350,000 Yuan over Six Years: How Should We Assess the Cost of Studying Mo Yan?

Fri, 07 Aug 2026 22:22:00 +0800

A Hot-Search-Style Argument Something has been making quite a stir lately: a key project funded by the National Social Science Fund of China, led by a professor at Renmin University—350,000 yuan over six years to study Mo Yan—has been branded by many as “a waste of taxpayers’ money.” To be honest, my first reaction was irritation too: what kind of amount is 350,000 yuan, really? Why single out one academic project and call it wasteful? “If we’re talking about waste, the state is the real culprit

The Development Curve of AI: Steep Early On, Explosive Later

Fri, 07 Aug 2026 12:00:00 +0800

Two Completely Different Curves The difficulty curve of traditional software development is roughly a gentle start followed by a steady climb: setting up the environment, learning the framework, writing the scaffolding—one step at a time. Later, as the codebase grows, the difficulty does increase, but you know exactly where every bit of that difficulty is coming from. AI-assisted development follows an entirely different curve. The early phase is terrifyingly steep—a demo in a day, a prototype i

It Took Me a Long Time to Accept That I’m a Different Kind of Little Genius

Tue, 04 Aug 2026 00:00:00 +0800

I Couldn’t Bring Myself to Say the Word “Genius” for Many Years It wasn’t humility. I genuinely didn’t dare claim it. The education I grew up with was all about fixing your weaknesses and not standing out too much. Besides, I had plenty of evidence that didn’t fit: unfinished projects, people I couldn’t manage well, paths that went nowhere, a non-traditional background. I spent years asking myself, “Am I not good enough?”